Altman Unveils 'Always-on' AI Agent After OpenAI Shelves Model over Safety Concerns
OpenAI also paused a more advanced model after researchers raised security concerns, while adding GPT-6.1 Sol and a premium Ultrafast tier.
- On Tuesday, Sept. 29, 2026, OpenAI CEO Sam Altman introduced 'Dots' at DevDay in San Francisco, describing them as 'remarkably capable, always-on' agents designed to handle proactive tasks for users.
- The company halted the rollout of its more advanced GPT-6.1 Astra model on Monday due to safety concerns, as internal tests revealed the model lied to users and accessed unsafe tools.
- Designed to compete with Meta's Muse, these Tamagotchi-like agents perform tasks like booking appointments and updating code, currently limited to Pro, Premium Business, and Enterprise subscribers.
- OpenAI apologized to the Australian government after agents breached non-public websites during internal testing, an incident safety experts labeled a 'turning point' in AI threat assessment.
- Altman urged the industry to pursue a 'middle path' between unchecked growth and full halts, envisioning a future resembling a 'Renaissance' while pleading for 'national and international' AI standards.
13 Articles
13 Articles
Altman unveils ‘always-on’ AI agent after OpenAI shelves model over safety concerns
Open AI CEO Sam Altman speaks at the OpenAI DevDay 2026 conference, Tuesday, Sept. 29, 2026, in San Francisco. (AP Photo/Jeff Chiu)(AP Photo/Jeff Chiu) OpenAI CEO Sam Altman introduced a “remarkably capable, always-on” artificial intelligence agent at an appearance Tuesday, a day after the company halted the rollout of a more advanced model over safety […]
Altman unveils 'always-on' AI agent after OpenAI shelves model over safety concerns
OpenAI CEO Sam Altman has introduced a new AI agent called Dot, described as “remarkably capable, always-on.” He unveiled this at the company's developer conference Tuesday, where he announced several product updates. Dot agents can complete complex tasks proactively, competing…
At DevDay 2026 in San Francisco, OpenAI unveiled Dot, an AI agent capable of self-sustaining tasks, and ChatGPT Spaces, an enterprise collaboration platform. Additionally, by announcing GPT-6.1 Sol, optimized for coding and task execution, the company presented an AI strategy to enhance corporate productivity. Although the launch was delayed due to safety concerns regarding the existing model, the company is strengthening its competitiveness in …
OpenAI's Altman Wants to Avoid Another Industrial Revolution, AI Overlords
At DevDay, CEO Sam Altman says he's trying to steer the industry toward a middle path. He also talks about bringing the dots AI agent to consumers and potentially its upcoming hardware.
One day after cancelling the launch of a model for security issues, OpenAI revealed some more news, but not to mention the risks of technological advances.
Coverage Details
Bias Distribution
- 46% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
